A total of 48 male C57BL/6J mice (RRID: IMSR_JAX:000664) aged 42 weeks (±3 days) were obtained from The Jackson Laboratory. In the present study, 42-week-old mice were selected because this age is equivalent to middle-age in humans (37,38). On the day of arrival, the mice were randomly placed in a cage, with two mice per cage. Placing two mice per cage has been shown previously to provide an essential environment for social interaction and tracking food consumption (39,40). Mice were housed in an AAALAC accredited animal facility under standard conditions: The ambient temperature 68-79˚F, relative humidity 30-70%, 10-15 air changes per h, and a 12/12-h light/dark cycle.

After 1 week of habituation, the 43-week-old mice (body weight 33.9±0.6 g) were administered either the AGE supplement or the control diet for 40 weeks (~10 months). Control mice were fed the nutritionally complete diet AIN93G (cat. no. F3156-Rodent Diet; AIN-93G Bio-Serv^®), and the AGE mice were fed the AIN diet supplemented with AGE (cat. no. F10217 Rodent Diet; AIN-93G-Modified; AGE-CSA-Missouri; Bio-Serv^®). AGE (40%; w/w aqueous solution) was provided by Wakunaga Pharmaceutical Co., Ltd. The AGE diet was prepared using the AIN93G base diet from Bio-Serv. AGE solution (2 kg) in 26 kg of the base diet (25.2 kg base diet powder and 2 kg AGE solution) was used to prepare 26 kg of the AGE diet. Because 2 kg of AGE solution contains 0.8 kg dry weight and 1.2 kg of water, the diet was dried at 120˚F in order to bring the moisture down to 5% after pelleting for microbial stability. For the control diet, 8% water was added to the dry mix as a pelleting aid. Mice were given the specific diets along with ad libitum access to water. Food intake was determined every week and body weights were measured every 2 weeks. To avoid additional stress, body weight was measured after behavior tests.

After feeding the mice the respective diets for 40 weeks, mice were subjected to behavior assessments. Prior to behavior tests, mice were housed in a reverse 12/12-h light/dark cycle environment for 2 weeks. All animal protocols were approved by the University of Missouri Animal Care and Use Committee (ACUC) (approval no. 25120; Columbia, USA).

The present study strictly adhered to the guidelines set forth by the University of Missouri and the National Institutes of Health for conducting animal studies and followed an approved method of euthanasia (The American Veterinary Medical Association-approved method). Mice should be euthanized before they experience significant pain or distress. The general appearance of the mice was checked by monitoring signs such as dehydration, severe weight loss, persistent abnormal posture or hypothermia, behavioral changes (such as inability to reach food or water, or lethargy) and clinical signs (such as labored breathing). The mice were euthanized with ~5% isoflurane and it was confirmed that there was no response to a firm toe pinch, and no signs of breathing and heartbeats, and the mice were continued to be exposed to ~5% isoflurane for ≥1 min to ensure euthanasia. Subsequently, the mice were perfused with saline through a cannula inserted into the left ventricle of the heart to wash out the blood. Removing blood from the brain results in cleaner tissue samples, which can enhance the accuracy and reliability of results. The mice were decapitated, and brain tissues were collected for future studies.

The aim of the battery of neurobehavioral tests was to assess the impact of dietary AGE supplementation on multiple behavioral domains associated with natural aging in mice (Fig. S1). All behavior tests were carried out in the dark phase (resting phase; with red lights in the testing room). Prior to the behavior tests, mice were placed in the testing room for ≥1 h for acclimation, and tests were conducted from 9:00 am to 3:30 pm. A battery of comprehensive behavior tests was established to evaluate different phenotypes under functional and cognitive behavioral domains, including sensorimotor function, exploratory behavior, anxiety/neophobia, memory and learning. To test sensorimotor functions, a simple neuro-assessment of asymmetric impairment (SNAP) test and an inverted screen test were performed. For anxiety/neophobia responses, the light-dark transition test, emergence test, open-field test, elevated-plus maze and zero maze were used. To assess memory and learning, the novel object recognition (NOR), three-chamber sociability and Barnes maze tests were performed.

The SNAP test was used to measure common neurological function domains including sensorimotor responses. The eight test phenotypes consist of interaction, cage grasp, visual placing, pacing/circling, gait/posture, head tilt, visual field and baton, as described previously (41). The SNAP test provides a relatively sensitive, reliable and time-efficient means of assessing neurological deficits in mice. Each individual test has a scoring range of 0-5 based on the established guidelines for atypical neurologic behavior (42).

The inverted screen test was used to measure motor function, coordination and muscle strength under the functional behavior domain. It is a quick but insensitive gross screening test that provides a measure of muscular strength. This test was performed as previously described by Deacon with modifications (43). Untrained mice were individually placed on top of a square (7.5x7.5 cm) wire screen which is mounted horizontally on a metal rod. Within 2 sec, the wire screen was rotated to an inverted position (180˚), displacing the mouse to the bottom of the screens. The mouse was kept in this position for 1 min (checking with a stopwatch). This position causes the mouse to fall. Normally, the mouse can hold the screen within a period of time and climb up the incline. The following parameters were measured: i) latency to fall off the screen, ii) could not climb up the screen, and iii) able to climb up the screen within 1-min testing session.

The light-dark box is a behavioral test that measures novelty-seeking and anxiety-like behaviors in mice under the cognitive domain. The test, performed according to the authors' previous studies (44,45), has the same area as the open-field test, except that half of the area is closed, and a small opening (7.5 cm high and 5 cm wide) could be found in the middle. Briefly, animals were placed in the light side of the box and were allowed to explore the box for 5 min. Their movements were tracked and analyzed using the ANY-maze software (RRID: SCR_014289; V7.16, Stoelting Co., Wood Dale, IL; Windows XP OS). A total of 3 measurements were calculated for each mouse: i) latency to first entry into the dark area, ii) number of entries to the dark area, and iii) time spent in the light area.

As described previously, the open-field test was performed to measure locomotion, exploratory activity and anxiety-like behaviors (44,46,47). The open-field activity box is 40x40 cm and located on a white platform. Animals were placed into the center of the square arena for 10 min. After each trial, the box was cleaned with 10% ethanol solution and air-dried. The animals' movements were tracked and analyzed using the ANY-maze software. A series of 10x10-cm blocks were identified to track mouse activity, with 4 blocks defined as the center zone and the surrounding 12 blocks as the peripheral zone. Phenotypic parameters such as total distance traveled and time spent in different zones (that is, center vs. periphery) were recorded as indicators of anxiety.

The NOR test in the open-field platform evaluates non-spatial learning and recognition memory under the cognition behavioral domain, as previously described (48), with a few modifications. First, the animal was allowed to become familiarized with an object for 5 min. After a 20~30 min interval, the object was placed in a different location in the same arena. Mice were allowed to explore the familiar object in the different location for 5 min. After 24 h, a new (novel) object was placed in the platform together with the familiar object. Normally, the color or shape of the novel object was changed, and the mouse was allowed to explore the object for 5 min. The entire test was recorded and measured using the ANY-maze software. In the aforementioned test, the following parameters were measured: i) Discrimination Index (DI) and ii) percentage of total investigating time. The DI was calculated by taking the difference between time investigating the novel object and the time investigating the familiar object, divided by the total time investigating both objects. The percentage of total investigating time was calculated by dividing the time investigating the novel object by the total time investigating both objects.

The emergence test is widely used to assess neophobia and exploratory behavior in rodents (49,50). The apparatus is an adaptation of the open-field test and intended to evaluate novelty-based anxiety. Naturally, rodents have an innate explorative drive as well as an aversion to brightly lit open spaces. The task involves observation of the time taken for the subject to exit the holding container (shelter) and explore the arena. Mice with high levels of anxiety tend to spend more time in shelter than in the open spaces.

Briefly, a large open field arena surrounded by high walls was used for this test. A cylindrical shaped tube (13.2x4.4x4.4 cm) that serves as the holding container or shelter, was used to restrain the mouse before placing in the arena. The shelter was equipped with lids and secured in place to prevent it from rolling in the center of the arena. While carrying the mouse in the tube, it was ensured not to close the lid tightly to maintain proper ventilation. After placing the tube in the center, the lid was removed. The number and time the subject spent in the shelter (tube), number of pokes into the shelter (tube) door, and time spent in the lit arena, were recorded and measured using the ANY-maze software.

The elevated plus maze test was used to evaluate the anxiety and fear domains. The test is based on the concept that rodent's natural behavior is to stay in the enclosed areas and display unconditional fear in open spaces and elevated heights. Anxious animals tend to spend more time in the closed arms than less anxious animals. The elevated plus maze was built according to the description of Lister (51). This maze was made of Plexiglas and consisted of a central square (5x5 cm) from which radiated four arms (5x45 cm). Two of the arms had walls (15 cm high) along the edge (closed arms), whereas the other two arms did not have walls (open arms). A white line was drawn halfway (22.5 cm) along each of the four arms to measure motor activity. The maze was elevated 45 cm above the floor on a plus-shaped plywood stand. The mouse was placed at the junction of the four arms of the maze, facing a closed arm. They were allowed to explore the apparatus for 5 min while an observer, a video camera, and the automated tracking system recorded their actions. Time spent in the open/closed arms and the duration of time in the center were tracked by ANY-maze software and used those for scoring various cognitive behavior parameters including: i) Number of entries in the open or closed arms, ii) time spent in the open or closed arms, iii) time spent in the center, and iv) percentage of time spent in the open/closed arms.

The zero maze was used to test anxiety- and exploratory-related behaviors in mice. It is an elevated, ring-shaped runway, with the same amount of area devoted to the adjacent open and closed quadrants. Mice were placed in one of the closed quadrants at the start of each trial and were allowed to explore the apparatus for 5 min. The ANY-maze software was used for scoring cognitive behavior parameters including: i) Number of entries in the open or closed segments, ii) time spent in the open or closed segments, iii) time spent in the center, and iv) percentage of time spent in the open/closed segments.

A three-chamber social interaction test was performed to assess sociability and social preference as described previously (52). Mice were habituated in the middle compartment for five min before testing. The ‘strange’ animal (same age and strain) was placed in the wire cage of either the left or right compartment. The strange mouse's location represented stranger zone 1 while the other wire cage remained empty (that is, empty zone). Mice were left to socialize for 10 min. The time spent in the stranger zone 1 (interaction with the new mouse) relative to the time spent in the empty zone was measured.

The social preference test was conducted for another 10 min after termination of the sociability test. A second strange mouse was introduced to the wire cage in place of the empty zone and was considered stranger zone 2. The animal's preference for interacting with a familiar animal (stranger zone 1) or an unfamiliar animal (stranger zone 2) was determined by measuring the same parameters. The indices of sociability and social preference were computed, as previously described (53).

Spatial learning and memory were tested using the Barnes maze as described (44,54). The maze consists of a circular platform (75 cm diameter) with 20 holes (5 cm diameter) evenly spaced around the platform's perimeter and is elevated 56.5 cm above the floor by a stand. One hole was designated as the escape hole for each mouse, and an escape box was placed underneath the hole. For each mouse, the location of the escape box remained the same throughout all test runs. To create an aversive environment, three 72-watt lights were suspended above the platform, encouraging the mice to flee from the highly lit environment into the darkened environment of the box. Four shapes (+, ∎, ∆, •) served as visual cues and each shape respectively was placed on each side of the black background curtains to aid spatial navigation. Behavioral testing consisted of 2 habituation trials on the first day, and 8 evaluation trials (2 trials/day) over the next 4 days. Reversal training consisted of 2 trials per day for 3 days, with an inter-trial interval of 20-30 min. This phase began 24 h after the final acquisition trial. Prior to the start of reversal training, each mouse was assigned a new escape hole, positioned opposite the previously assigned location. Latency (that is, the time mice found the escape box) and total errors (that is, nose-pokes into non-escape holes) were recorded and analyzed by ANY-maze software.

The data are presented as the mean ± SEM. Results were analyzed by one-way or two-way ANOVA using GraphPad Prism Software (RRID:SCR_002798; V10.0; Dotmatics). Tukey's post hoc test was applied for multiple comparisons. Differences were considered statistically significant at P<0.05 for all analyses. An unpaired t-test was applied for individual comparisons between two groups.

A total of 10 mice, AGE (n=5) and control diet (n=5), were euthanized at 88 weeks-old (following 11 months of AGE supplementation) for label-free global proteomic quantitation by tandem mass spectrometry (MS²) analysis. The sample preparation for quantitative proteomic analysis has been described previously (55). Briefly, the left cortex and hippocampi from both hemispheres were dissected and processed. Tissues were lysed with sample buffer containing 2% sodium dodecyl sulphate (SDS), 0.5 M tetraethylammonium bicarbonate (TEAB), pH 8.5, protease inhibitor cocktail (1:100), and phosphatase inhibitor cocktail (1:100). Specimens were homogenized by Glas-Col stringer 099C K43 (Glas-Col LLC, IN) and centrifuged at 17,000 x g for 20 min at 4˚C. The protein pellets (insoluble fraction) were collected and further precipitated with 4X (w/v) acetone and incubated at -20˚C overnight. Prior to MS² analysis, protein pellets were washed two more times with acetone.

Proteins in samples were denatured with 6 M urea, 2 M thiourea and 100 mM ammonium bicarbonate, and then reduced with 10 mM dithiothreitol (DTT) for 1 h at room temperature (RT) and alkylated at RT in the dark with 30 mM iodoacetamide. Alkylation was halted with 10 mM DTT for 15 min at RT. Protein concentration was determined using Pierce 660 nm Protein Assay (Thermo Fisher Scientific, Inc.) according to the manufacturer's protocol. Proteins from each sample were digested with trypsin (Thermo Fisher Scientific, Inc.) in a sequential digestion. Trypsin was added at a 1:100 enzyme to protein mass ratio. The first digest was incubated at 37˚C for 5 h, and the second digest was incubated overnight. The digested peptides were terminated with trifluoroacetic acid (TFA) 0.5%. For total protein quantification, 5-µg of peptide from each sample was combined for spectral library using the method for data-dependent acquisition with parallel accumulation-serial fragmentation (DDA-PASEF). Subsequently, 500 ng peptides from each sample were loaded to the liquid chromatography-MS and analyzed using the method for data-independent acquisition (DIA-PASEF).

To create a comprehensive spectral library, digested peptides were combined from each sample and fractionated using a high pH reversed-phase peptide fractionation kit according to the manufacturer's protocol (Thermo Fisher Scientific, Inc.). A total of eight fractions were collected, lyophilized and resuspended with 0.1% formic acid in deionized water.

For proteome analyses, an EvoSep One liquid chromatography system was used as described previously (56) and the peptides were analyzed by eluting with a 44 min gradient or EvoSep One 30 samples per day (SPD) program. The system used a 15 cm x 150 µm ID column with 1.5 µm C18 beads (Bruker PepSep) and a 20 µm ID zero dead volume electrospray emitter (Bruker Daltonics). Mobile phases A and B were 0.1% formic acid in water and 0.1% formic acid in acetonitrile, respectively. EvoSep One was coupled online to a modified trapped ion mobility spectrometry quadrupole time-of-flight mass spectrometer (timsTOF Pro 2, Bruker Daltonics) via a nano electrospray ion source (Captivespray, Bruker Daltonics). To generate a spectral library, eight high-pH reversed-phase fractionated samples were analyzed by timsTOF Pro 2 in the DDA-PASEF mode. PASEF and TIMS were set to ‘on’. One MS and ten PASEF frames were acquired per cycle of 1.17 sec (~1 MS and 120 MS²). Target intensity for MS was set at 10,000 counts/sec with a minimum threshold of 2,500 counts/sec. Duty cycle was locked to 100%. Ion mobility coefficient (1/K0) value was set from 0.6 to1.6 V.s/cm2, collision energy was set from 20-59 eV. MS data were collected over m/z range of 100 to 1,700. If the precursor (within mass width error of 0.015 m/z) had >4X signal intensity in subsequent scans, a second MS² spectrum was collected. Exclusion was active after 0.4 min. Isolation width was set to 2 for m/z <700 m/z or 3 m/z for m/z >700.

The DIA-PASEF method was optimized with py_diAID, as described previously (57), to cover an m/z range from 300 to 1,350 Da and an ion mobility range of 0.65 to 1.45 V.s/cm2. The method includes two IM windows per DIA-PASEF scan with variable isolation window widths adjusted to the precursor densities. A total of 25 DIA-PASEF scans were deployed at a throughput of 30 SPDs (cycle time: 2.76 sec).

Data were analyzed with Spectronaut version 17 (Biognosys AG) with the default settings, except for the prototypicity filter, which was set to ‘only protein group specific’. Protein quantification and statistical analysis were performed using MSstats. The MSstat set up included protein quantification using the top 3 features, a q-value cutoff of 0.01, the use of unique peptides, and the removal of protein with only one feature. No imputation was used, and the data were log₂-transfered based on MS² peak area. Normalization was performed using the ‘equalize medians’ method.

AGE supplementation-induced molecular phenome changes in SDS-insoluble protein were evaluated by computing the expression ratio and P-value for all MS²-identified proteins from AGE mice relative to age-matched controls. The expression ratio threshold was set to >1.3 and <0.77 for respective increased and decreased proteins (P<0.05). For data visualization, expression ratios and p values were log-transformed to produce fold-change (log₂ expression ratio) and -log₁₀ P-value.

The differentially changed proteins in AGE mice cortex and hippocampus were imported into Ingenuity Pathway Analysis (IPA; Redwood City, CA; RRID:SCR_008653) for canonical pathway analysis, machine learning-driven disease pathway analysis, upstream regulator analysis, and neurological behavior prediction. Additionally, comparative analysis was performed by combining expression profiles from AGE mice cortex and hippocampus to obtain 2x2 comparisons for top pathways. The data are licensed under the Creative Commons Attribution-NonCommercial 4.0 International License (CC BY-NC 4.0).

No significant differences in average food intake or body weight (Figs. S2 and S3, respectively) were observed when comparing the AGE-supplemented diet-fed mice with the control diet-fed mice. Additionally, no changes in mortality or apparent abnormalities in general health were found in the mice during the 40-week feeding period. Neither group exhibited any significant decrease in body weight throughout the study. The maximum observed body weight loss occurred at the 34th feeding-week, with a weight loss of ~8.1% in the AGE diet-fed group and 6.0% in the control diet-fed group compared with the 1st feeding-week.

The SNAP test was used to assess the impact of AGE supplementation on the sensorimotor function of the aging mice. The inverted screen grip test was performed to assess the effects of AGE on physical strength. No statistically significant differences in coordination, motor strength, proprioception, gait and mental status were observed following 40-week dietary AGE supplementation (Fig. S4).

Mice fed the AGE diet exhibited less anxiety in the light-dark transition test and reduced neophobia-like behaviors in the emergence test. These mice exhibited a decreased number of entries (Fig. 1A; P<0.01) and time spent in the dark zone (Fig. 1B; P=0.046) relative to the aging controls. Furthermore, the AGE group exhibited an increased number of entries (Fig. 1C; P<0.05) and time spent in the light zone (Fig. 1D; P=0.034).

The AGE-supplemented diet-fed group also exhibited less anxiety and neophobia and increased exploratory behavior in the emergence test (Fig. 2A-G), with statistically significant differences in the following parameters: Less time spent in the shelter zone, more time spent in the arena zone, higher number of line crossings in the arena zone, longer distance travelled in the arena and less time immobile in the arena zone. However, no differences were observed between the AGE-supplemented and control diet groups in the open field test (Fig. S5A-C), as well as the elevated plus maze and zero maze tests (Fig. S6A-F).

In the NOR test to evaluate ability of non-spatial learning and recognition memory under the cognition behavioral domain, AGE-supplemented mice displayed a greater DI (Fig. 3A; P=0.027), and increased percentage of investigation time with the novel object location (Fig. 3B; P=0.039), indicating AGE supplementation could improve learning levels, memory and exploratory behaviors in aging mice.

It was assessed if AGE supplementation could improve spatial reference learning abilities in aging mice. During the 5-min Barnes maze test, a higher percentage of mice in the AGE diet-fed group found the hiding box for acquired learning compared with that in the control group. For those that found within the allotted time, the latency to enter the hiding box was used to estimate memory and learning abilities. In reversal training, AGE-supplemented diet-fed mice displayed a shorter latency to enter the target zone compared with mice in the control group. The differences were most pronounced in trial 1 on day 2 of reversal training (Fig. 4).

The present study evaluated whether AGE supplementation could improve social recognition, sociability and novelty preference behavior. The three-chamber social interaction test was used for evaluating impact of AGE on the social index of aging mice (time spent with stranger mouse 1 vs. empty cages) and on the novelty preference index [time spent with familiar (stranger mouse 1) vs. stranger mouse 2]. In the present study, AGE supplementation did not improve cognitive domain parameters (Fig. S7A and B) under the current conditions.

The brain proteomic profiles were examined to identify the molecular effects of AGE supplementation in aging mice. Of 5,863 and 5,826 proteins identified using label-free MS², 90 and 336 proteins were significantly changed in the AGE-supplemented mouse cortex and hippocampus, respectively (Fig. 5A and B; Table SI). A total of 34 and 197 proteins were decreased in the cortex and hippocampus, respectively, whereas 139 and 56 proteins were increased in the cortex and hippocampus, respectively. Notably, dietary AGE supplementation decreased selenium-binding protein 1 (Selenbp1; fold-change in the cortex, -0.700; P=2.156x10^-5; fold-change in the hippocampus, -1.039; P=3.415x10^-5) and increased prodynorphin (Pdyn; fold-change in the cortex, 0.734; P=0.007; fold-change in the hippocampus, 1.035; P=0.010) in both brain regions (Fig. 5C).

With the bioinformatics tool to examine mouse phenomes, which refers to a full set of observable traits (for example, behavior, physiology and/or disease susceptibility), IPA was able to use curated proteomic datasets to forecast changes in phenotypic outcomes. IPA comparative analysis showed that molecular pathways in both the cortex and hippocampal regions were similarly impacted but with some important distinctions. The commonly affected pathways were ‘epithelial adherens junction signaling’ (increased; P<0.05), ‘synaptogenesis signaling pathway’ (increased; P<0.05) and ‘Sertoli cell-Sertoli cell junction signaling’ (increased; P<0.05) (Fig. 6A; Table SII, sheet: Canonical pathways). Among the top canonical pathways, 14-3-3-mediated signaling related to apoptosis was impacted exclusively in the AGE-supplemented hippocampus (z-score, -1.387; P=1.362x10^-5) (Fig. 6A; Table SII, sheet: Canonical pathways). IPA upstream regulator analysis predicted microtubule-associated protein tau (MAPT) to be most impacted by dietary AGE supplementation based on P-value ranking, and this occurred mostly in the hippocampus (cortex P=0.007; hippocampus P=1.212x10^-14) (Fig. 6B; Table SII, sheet: Upstream regulators). Amyloid precursor protein (APP) and brain-derived neurotrophic factor (BDNF) were predicted to be increased in the cortex and reduced in the hippocampus (Fig. 6B; Table SII, sheet: Upstream regulators). Presenilin 1 (PSEN1) was exclusively increased in the hippocampus (z-score, 0.658; P=1.992x10^-10) (Fig. 6B; Table SII, sheet: Upstream regulators). The p53 cell-cycle regulator was reduced in both the cortex (z-score, -0.442; P=0.002) and hippocampus (z-score, -1.781; P=1.917x10^-8) (Fig. 6B; Table SII, sheet: Upstream regulators). IPA behavior predictions suggested that differential expression changes in the cortex and hippocampus proteomes due to AGE supplementation could impact cognition and learning, whereas only changes to the hippocampus proteomes could impact anxiety, memory and spatial memory (Fig. 6C; Table SII, sheet: Behavior prediction).